BACKGROUND--Coexisting cardiorespiratory disease may contribute in individual cases to the increased ventilatory response to exercise in patients with chronic heart failure. OBJECTIVE--To characterise further the arterial blood gas response to exercise and to explore the possible uses of blood gas sampling in clinical practice in patients with chronic heart failure. METHODS--37 patients with a primary diagnosis of chronic heart failure (age (range) 59 (45-80); left ventricular ejection fraction 24.5% (4%-44%)) underwent exercise testing with arterial blood gas analysis during exercise. RESULTS--In 34 patients there was a small fall in arterial carbon dioxide tension from a mean (SEM) of 4.9 (0.1) kPa at rest to 4.6 (0.1) kPa at peak exercise (p < 0.001). There was no significant change in arterial oxygen tension. During the recovery period arterial oxygen tension rose from 13.3 (0.3) kPa at peak exercise to 14.8 (0.3) kPa three minutes into recovery (p < 0.001). Arterial carbon dioxide tension was unchanged. In the remaining three patients there was considerable arterial hypoxaemia on exercise, from 10.4 kPa at rest to 7.7 kPa at peak exercise. All of these patients had an alternative diagnosis (patent foramen ovale with right to left shunt during exercise, pulmonary embolic disease, and clinically unsuspected obstructive airways disease). CONCLUSION--Patients with a presumptive diagnosis of chronic heart failure should undergo exercise testing with arterial blood gas analysis. Arterial hypoxaemia on exercise is rare in stable chronic heart failure. If hypoxia on exercise is detected, an alternative diagnosis should be sought.